Ultra-low ESL capacitors driving coils / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Ultra-low ESL capacitors driving coils

Move Thread

LAN_403

Signification

Sun Feb 15 2015, 11:15AM

Registered Member #54278 Joined: Sat Jan 17 2015, 04:42AM
Location: Amite, La.
Posts: 367

I have seen it stated repeatedly, in projects that drive coils (inductive loads), such as can crushers and coin shrinkers, that capacitors with ultra low ESL (Equivalent Series Inductance) must be used. WHY? If the capacitor is driving an inductive load anyway, What difference is a few nano-Henries of internal inductance in the capacitor going to make?

DerAlbi

Sun Feb 15 2015, 09:08PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Sure thats not asked in some old thread you could dig out from years ago by adding some random information?

but to give an answer: it depends in your topology.
if you use ZCS switching its not worth thinking about.

if you use hard (or ZVS) switching, then the halfbridge / fullbridge will suffer from every additinal nH by disipating heat as switching loss. When it gets to higher Powers this is indeed a big problem.
This of course, applies only to the DC-decoupling capacitor (and Halfbridge routing). Any seriescapacitance to an inductor does not need to be selected by ESL.

Dr. Slack

Sun Feb 15 2015, 11:50PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Signification wrote ...

I have seen it stated repeatedly, in projects that drive coils (inductive loads), such as can crushers and coin shrinkers, that capacitors with ultra low ESL (Equivalent Series Inductance) must be used. WHY? If the capacitor is driving an inductive load anyway, What difference is a few nano-Henries of internal inductance in the capacitor going to make?

Because if you make a crappy cap bank with electrolytics, you're not going to get a few nano-Hs, you're going to an order or two of magnitude more. I made a 4kV 50uF bank from 100uF 330V photoflash caps ( a lot of soldering!), and used a 5kv 50uF pulse cap charged to 4kv. One would neck a can down to the width of a finger, the other would put very tiny, delicate ripples into its surface. Guess which was which.

Hint, even though you might calculate the inductance of your work coil in fresh air, when it is coupled to an alli can, the inductance is reduced radically by the can's shorted turn.

Now, if you want to drive a disc launcher, you *can* ride out the excess micro-Hs without much ill effect.

DerAlbi

Mon Feb 16 2015, 04:41PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Your ESL was not necessarily composed of the added ESLs of your Caps, but the fact that you build up such a capactior(bank) in an unappropiate way. The size/length of a cpacitor string alone has inherent inductance so that it can have huge effect.
In series connected capacitors have no solwer current rise time than the single cap.

klugesmith

Mon Feb 16 2015, 06:05PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

We agree that the ESL of an individual capacitor sets a practical limit on the discharge rate.
The associated time constant, sqrt(LC), is the same for any bank of identical cells in series and/or parallel,
before we consider the interconnection inductance.

As you suggest, ignorant designers could get it wrong there.
I bet Dr Slack has what it takes to construct a bank not much slower than the individual capacitors.
And concur that aluminum electrolytic capacitors are inherently slow for can crushing (but OK for coilguns).

My own 52 uF HV capacitor, at 4 kV, crushed cans about as well as Dr Slack's. It resonated at about 15 kHz with empty coil. If I recall correctly, its measured self-resonant frequency was about 100 kHz -- and of course frequency ratio is the square root of inductance ratio. Does anybody here happen to know the SRF of "low ESL" energy storage electrolytics?

The practical speed limit of a capacitor bank can be determined with an oscilloscope. With bank charged to a relatively tiny voltage, measure current pulse as bank is short-circuited. (e.g. by connecting to a current shunt). The inductance of regular power resistors in the sub-ohm range is negligible at 100 kHz.

DerAlbi

Tue Feb 17 2015, 02:33AM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

SRF figures should be seen in a datasheet. There is no absolute statement to this. it depends on manufacturer and model/series and the capacitance of course - not every cap benefits from multiple taps on the foil equally.. You need to ask more specific in that case.

Concerining the design of a cap bank.. i really dont know if its that obvious to everyone that its beneficial to connect the paralleled caps in series instead of connecting the series caps in parallel.

Signification

Tue Feb 17 2015, 04:03PM

Registered Member #54278 Joined: Sat Jan 17 2015, 04:42AM
Location: Amite, La.
Posts: 367

Dr.Slack makes clear the significance of capacitor ESL by pointing out how critical this ESL is in a 'can crusher' capacitor due to the "shunted" inductance the work coil experiences due to the Al can acting as a shorted turn.
In addition he states, as does Klugesmith, that 'non-low' ESL capacitors i.e. electrolytics are acceptable in a disc launcher. But since the disc in the disc launcher is also a 'shorted turn' I am assuming this is an 'acceptable' application due to the fact that the disc quickly jumps out of the field. BUT, wasn't the disc, in this case, acting as the same 'shorted turn' during the fast but critical 'launch pulse' (call it tp)? Is the explanation that a much SMALLER tp is required for a 'good' disc launch OR a LARGER tp required for a 'good' crush on the can crusher?

I think it would be very informative, and would like to hear other members' opinions on the consideration of a third case with comparsions on the 'goal requirements' of each. A third case is that of a 'coin-shrinker'. In this case the shorted turn (the coin), as with the can-crusher remains. I would assume a very low ESL is required here, as the crushing force is caused by fast current rise-time (di/dt) in the working coil (much more, due to transformer action, is circulating in the coin). Since such an incredibly high coil di/dt is required to shrink a quarter and the work coil can sustain it, and the simultaneous opposite force (Lenz), for only a few 10's of us. I assume two things are required:
1) An energy capacitor with a low ESL since the coin is a shorted turn.
2) A very small tp requirement since the source of the shrinking force (the coil's di/dt) is quickly decoupled from the coin (similar to the disc-launcher) but by explosive destruction of the coil rather than separation.

DerAlbi makes a statement that is desirable but very counter-intuitive: "In series connected capacitors have no slower current rise time than the single cap." But after considering an illustrative situation carefully, makes sense: In a drawing of, for example, three series-connected capacitors with each component drawn as a series-connected capacitor (C) + resistor (R) we see that the resistance is tripled, but also illustrated is three capacitors connected in series giving an equivalent C/3. This shows a 3R : C/3 component. That an applied voltage source, for example, causes a current to encounter three times the resistance but need only charge 1/3 the capacitance. Better, Looking at things from a resistor-capacitor 'time-constant' point of view: the RC time constant is 3R*(C/3) = RC: the same as that of a single capacitor.

DerAlbi also says: "...its beneficial to connect the paralleled caps in series instead of connecting the series caps in parallel." Can you be more specific here as to the particular advantage you speak of? I assume you are speaking of connecting that 'third' center node of the "series caps connected in parallel". In my work with capacitor bank configurations for rapid-fire flashlamps of pulsed lasers, it is recommended to NOT connect this node. As in this configuration, if one capacitor fails shorted, then all other 'banks' are over-volted.

Finally, I have always assumed if capacitors of the same type and rating are connected in series then the ESR increases the same as resistors in series--This, I still do believe. Is there any different opinions? BUT, concerning capacitors in parallel, I no longer think the ESR DECREASES the same as resistors in parallel. So, is this correct? If so how is total ESR related to capacitors in parallel??

hen918

Tue Feb 17 2015, 07:21PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

Signification wrote ...

Dr.Slack makes clear the significance of capacitor ESL by pointing out how critical this ESL is in a 'can crusher' capacitor due to the "shunted" inductance the work coil experiences due to the Al can acting as a shorted turn.
In addition he states, as does Klugesmith, that 'non-low' ESL capacitors i.e. electrolytics are acceptable in a disc launcher. But since the disc in the disc launcher is also a 'shorted turn' I am assuming this is an 'acceptable' application due to the fact that the disc quickly jumps out of the field. BUT, wasn't the disc, in this case, acting as the same 'shorted turn' during the fast but critical 'launch pulse' (call it tp)? Is the explanation that a much SMALLER tp is required for a 'good' disc launch OR a LARGER tp required for a 'good' crush on the can crusher?

...

I also noticed that, and assumed that he meant there are usually more turns on a disk launcher and therefore more inductance.

klugesmith

Tue Feb 17 2015, 09:24PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

OK, let's see some pulse width numbers.

The best coil in my can-crushing experiments gave PW of around 25 microseconds.
4hv reader tesladownunder reported about the same in 2004.

(It was he, Peter, who referred me to 4hv to talk about it.)

For coin shrinking, using coils with smaller diameter but more turns, Hackerbotlabs presented video evidence of 30 - 40 us dynamics. Maybe Bert Hickman's website mentions some of his measurements.

I think the PW in amateur coilguns is measured in milliseconds.

Don't know about disk launchers. Who wants to chime in about them?

If anyone is serious about undertaking new measurements, I can provide some of my printed-circuit-board Rogowski coils, whose sensitivity (mutual inductance) is (IIRC) around 35 nH.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob